Reaping the Rewards of Resource Consolidation
The impact of a sophisticated service level management solution on resource consolidation initiatives

by Christopher Marino


Falling Short of the Million Dollar Promise

Saving millions of dollars from a resource consolidation initiative is not as simple as replacing hundreds of little servers with a few big ones. Failure to implement appropriate management software as part of this type of initiative can limit the expected savings. Worse yet, without sophisticated management software capabilities in place, a decrease in performance and availability can be the price that is paid for saving a few dollars in hardware.

Over Provisioning: A Thing of the Past

With IT budgets tighter than ever, gone are the days when companies can afford to deploy an overcapacity of servers in anticipation of a stampede of online customers. IT managers are being asked to accomplish more with less  less people, less servers, less software, less consulting. No longer is it acceptable to run servers at 10-20% utilization, a common paractice in the recent past. Server utilization is now the focus for increasing productivity to reduce costs.

Addressing End-User Needs and Business Expectations

Unfortunately for IT managers, doing more with less does not mean that executives and business units are willing to lower performance and availability expectations for business-critical customer services. In fact, the opposite is true. Executives today expect improvements in the performance and availability of existing IP-based services, while also expecting reductions in operating and capital costs.

Resource Consolidation Options

There are several ways for companies to improve resource utilization and reduce server costs. One is centralization, the moving of servers distributed across an enterprise to large data centers to improve the costs of management. Another is server consolidation, the reduction of many smaller capacity servers into a few high capacity servers.

The first step in approaching a consolidation project is to evaluate the existing infrastructure. What can be done with current resources to maintain or improve service levels without incurring additional high cost expenses? Application stacking and running servers at higher capacities are two such options to consider. Application stacking reduces server costs by stacking,  or running, several applications previously on multiple servers, on to fewer servers that were previously underutilized. In addition, servers can be run at higher capacities, or closer to the red line  for better resource utilization.

None of the above approaches are mutually exclusive, and in fact often a combination of solutions are implemented in a consolidation project.

Improving Service Levels While Consolidating Resources

Essential to the success of any consolidation project, is the foresight to implement an appropriate service level management approach. Deploying a sophisticated and complete service level management (SLM) solution is essential to maintaining a positive end-user experience while reducing operating expenses. The benefits of implementing an SLM solution are more than just financial, and include increased capacity utilization, faster problem diagnosis, greater flexibility for rapid growth, and improved ROI.

Application Stacking to Improve Utilization

Stacking of applications, or running multiple applications on the same server which previously ran a single application, is a way to improve utilization of servers. Financial gains from successful stacking initiatives can look great on paper. For example, if 50 servers are running a customer account management application and 50 are running a mail application, stacking both on 25 servers can save $500K.

However, management challenges increase dramatically in a stacked application environment making a more sophisticated management solution a pre-requisite for success, and for avoiding disaster. See below for business-critical issues to avoid when application stacking (Table 1).

Management Issue

Management Software Requirement

Inter-application Performance Conflicts

Applications may adversely affect the performance of each other by bottlenecking on resources such as disk/io, memory, or network interfaces, requiring more sophisticated management software to analyze performance

Application Diagnosis

Diagnosing application problems is more complex, requiring tools to separate applications and correlate to other infrastructure components such as databases

Server Maintenance

Maintenance on servers requires more planning because more applications are affected by server maintenance operations

Table 1. Application Stacking Issues

 

Improving Utilization While Managing the Existing Infrastructure

Enterprises can realize tremendous gains from even small improvements in resource utilization levels. For example, running servers at an average capacity utilization of 20% is not uncommon. Increasing this to 30%, when applied across hundreds of servers, can save millions of dollars. For example, at an average cost of $20,000 per server (including management costs), a 200 server environment can realize savings of $1.3 million by increasing utilization from 20% to 30%. (Examples of the ROI of such initiatives are given later in this paper).

However, in order to ensure that client performance and availability are not sacrificed to achieve these gains, sophisticated management software is required. The issues at hand are summarized below (Table 2).

Management Issue

Management Software Requirement

Handling Spikes in Traffic and Computational Demands

Higher usage levels of critical server, application, and network resources require real-time visibility into performance metrics

Increased Exposure to Server Outage Impact

End-to-end visibility from client to network, applications, databases enables rapid location and diagnosis of performance and availability problems

Table 2. Redline Issues

Provisioning Promises

Recently, announcements have been made from major systems vendors about the concept of automated server and application provisioning. Successful development of such capabilities by vendors could substantially increase the gains from consolidation. These systems (and accompanying software), can dynamically allocate server resources to services experiencing high demand. This in turn enables enterprises to run many applications across a shared pool of server resources, saving time and money.

Several key capabilities, which are required building blocks for more advanced provisioning systems, exist today in several specialized products. These include:

  • Traffic management: The ability to instantly and dynamically route client connections to appropriate servers is required in environments where a server may be running one application this hour but another the next hour.
  • Quality of Service (QoS): During periods of resource shortages, a policy-based system for allocating resources according to the quality of service desired for different clients is required.
  • Real-time service level monitoring: Decisions about when to provision more resources or decrease resources to a particular service or application are based upon accurate measurements of service levels, measured externally for clients as well as internally for infrastructure components.

Implementation of a sophisticated, end-to-end SLM solution will prepare enterprises for future complex provisioning environments.

Returns from Server Consolidation: The Bottom Line

The financial returns from a successful consolidation can be tremendous. In the example below, a 100 server environment is projected to be able to save over $634 million in the first year by improving server efficiency by 33% (from 30% utilization to 40%). This is based on an assumption of $380,000 of upfront software costs to enable the company to achieve the maximum increases in utilization possible, as well as to protect client performance and availability. It also assumes an average costs per server of $20,000. Higher-end server consolidation projects would have a higher savings.

 

Year 0

Year 1

Year 2

Year 3

"Cash" Cost of Ownership

$380,000

$60,000

$60,000

$60,000

"Cash" Savings

$0

$634,615

$165,000

$165,000

Cu Cumulative Cost

$380,000

$440,000

$500,000

$560,000

Cumu Cumulative Savings

$0

$634,615

$799,615

$964,615

ROI

 

51%

68%

76%

Example 1. 100 Server ROI from 33% Improvement

The return on investment (ROI) from such a project can exceed 50% in the first year, and the payback of software and implementation costs can be achieved in six to nine months, as shown in the chart below (Figure 1).

Figure 1. Cumulative Costs vs. Savings

A complete and sophisticated service level management solution would not only enable savings from server infrastructure consolidations, but would also enable savings in labor and reduced downtimes. Reductions in the time it takes to diagnose problems can translate into cash savings from decreased labor costs for managing the service.

In the above example, if a 20% reduction in time to problem resolution, a 10% overall improvement in labor efficiency, and a 1% improvement in uptime are factored in, the ROI more than doubles to 125% in Year 1, and the payback period is reduced to about six months. Those are compelling numbers that can t be ignored by business and IT executives.

Best of Both Worlds: Financial Savings and Better Utilization

In addition to the financial benefits of deploying a complete and sophisticated SLM solution for server consolidation initiatives, there are many other benefits. These include greater resource utilization, faster problem diagnosis, and improved visibility and reporting. Additional benefits are summarized below (Table 3).

Benefit

Description

Increased Capacity Utilization

Higher usage levels of critical server, application, and network resources enabled by real-time visibility into performance metrics

Faster Problem Diagnosis

End-to-end visibility from client to network, applications, databases enables rapid location and diagnosis of performance and availability problems

Faster Problem Resolution

An active  solution enables rapid manual and automated resolution by dynamically adjusting critical performance parameters and restarting application components and processes

Improved Client Service Levels

Result from improved problem resolution and are enhanced by ability to route traffic around problem areas through integration with leading traffic management devices

Improved Management Visibility and Reporting

Setting, tracking, and reporting on Service Level Objectives (SLO s) and correlation to infrastructure bottlenecks enables management to address critical issues

Table 3. Additional Benefits

These benefits ensure that a consolidation project achieves goals that are consistent with the business objectives of the company, including financial, customer service, and IT operations goals.

Reaping the Benefits of Consolidation Initiatives

Companies seeking to get the most from large capital investments of recent years can look to consolidation projects for cost savings and capacity utilization improvements.

Valuable benefits including significantly higher ROI can be realized through successful deployment of a consolidation initiative. A successful initiative requires deployment of a sophisticated and complete service level management solution for managing performance and availability, enabling maximum returns on resource investments.


Appendix: Resource Consolidation and Related Terms

Server Consolidation: Reducing the number of servers required to run the same number of application(s).

Centralization: Moving distributed servers into data center(s), often resulting in consolidation opportunities.

Application: Stacking Running of several applications, often simultaneously, on a single server/operating system.

Capacity Utilization: Average capacity of server usage, typically represented by CPU load.

Provisioning: Dynamically allocating resources (server, application) to client requests optimizing price/performance levels.

Service Level Management: Managing the performance and availability of service levels including Service Level Objectives (SLO's) reporting to diagnose and resolve problems.

Quality of Service: Allocation of scarce resources to improve, and to some extent, guarantee the information being requested.


Christopher Marino brings over 18 years experience designing, implementing and marketing hardware and software systems to his role as founder, vice president and director at Resonate. Prior to founding Resonate, Mr. Marino was the Director of Product Marketing at nCUBE, a manufacturer of parallel-processing video server systems. Prior to nCUBE, Mr. Marino was product line manager at MIPS Computer Systems where he was responsible for the company's line of high-end multiprocessor server systems. Mr. Marino also held design-engineering positions at Zycad and Bell Laboratories. Mr. Marino holds a BS EE from Columbia University, an MS EE/CS from the University of California at Berkeley and an MBA from Stanford University. Mr. Marino is a member of Resonate's Board of Directors and serves as a Director for several private companies.

Copyright (c) 2000-2003, nextslm.org. All Rights Reserved. Legal Statement.